New persona
A fictional adult. Write the look as concrete photographic facts — Z-Image's text encoder reads prose well, and the candidates step renders straight from this text.
Identity
Look
Style & voice
What happens next
- 1. Render portrait candidates from this text and pick her face.
- 2. That face becomes the reference; refine it if needed.
- 3. Render the training set and train the identity LoRA.
- 4. Generate with the LoRA — same woman in every image.
Generation settings start from the Z-Image-Turbo defaults; the NSFW LoRA and any extra models are added on the persona page.